摘要 |
PROBLEM TO BE SOLVED: To provide a method for discriminating an object with less computational complexity and detecting its position and posture even when the object can be detected with high robustness regardless of characteristics of the objects and the objects are piled up. SOLUTION: An area which can be approximated to a plane surface of an object is extracted from distance data 101 of an object to be detected (102), its plane surface is calculated (103), the distance data are projected onto the calculated plane surface to generate a distance image (104), and this is compared (106) with a reference distance image of the object to be detected stored in a database (105) beforehand. Thus, a position and posture of the object are specified (107), to determine the position and posture of the object to be detected. COPYRIGHT: (C)2008,JPO&INPIT
|